Skip to main content

Epigenetics For the Spirit

You may have come across Siddhartha Mukherjee’s article in the New Yorker on epigenetics, and if so, you may also have noticed that it didn’t make a big hit with biologists who are actually working in the field. Here’s Mark Ptashne giving chapter and verse: the short version is that Mukherjee’s article lets people who don’t know the area (which takes in the vast majority of the people who will read it) come away with a pretty mistaken idea of what epigenetics really is. Particularly upsetting to many biologists was the way the article never even mentioned the phrase “transcription factors”, when the main function of all these epigenetic markers is, in fact, to harness and guide such proteins. The article centers around the differences between Mukherjee’s mother and her identical twin, which is fine as a starting point for discussions of heredity versus environment and how gene expression can be altered by environmental factors. But lay readers could well get the idea that epigenetic markers (such as histone methylation) are some sort of alternate genetic code that transmits information from generation to generation, and that’s not as clear-cut as the article makes it sound. The article certainly raises the possibility, and critics don’t think it does enough to show how contentious the evidence is:

Both Allis and Reinberg understand the implications of transgenerational epigenetic transmission: it would overturn fundamental principles of biology, including our understanding of evolution. Conceptually, a key element of classical Darwinian evolution is that genes do not retain an organism’s experiences in a permanently heritable manner. Jean-Baptiste Lamarck, in the early nineteenth century, had supposed that when an antelope strained its neck to reach a tree its efforts were somehow passed down and its progeny evolved into giraffes. Darwin discredited that model. Giraffes, he proposed, arose through heritable variation and natural selection—a tall-necked specimen appears in an ancestral tree-grazing animal, and, perhaps during a period of famine, this mutant survives and is naturally selected. But, if epigenetic information can be transmitted through sperm and eggs, an organism would seem to have a direct conduit to the heritable features of its progeny. Such a system would act as a wormhole for evolution—a shortcut through the glum cycles of mutation and natural selection.

I should note that Mukherjee has some replies to his critics here, talking with Matthew Herper. From a drug discovery standpoint, epigenetic targets have a lot of appeal because affecting transcription factors directly is brutally hard. So if you can inhibit an enzyme that modifies transcription downstream, you have a handle, because we can generally inhibit enzymes. There is that little problem that we don’t really understand what all these epigenetic enzymes are doing in that great big downstream of gene regulation, but having compounds that affect them is one good way to find out.

Michael Eisen has a good piece on this as well. And he makes a larger point that’s worth keeping in mind, that the word “epigenetics” isn’t just being abused in the wider press. It’s been abused by scientists as well:

But Mukherjee is far from the only one to have fallen into this trap. Which brings me to what I think is the most interesting question here: why does this particular type of epigenetic inheritance involving an obscure biochemical process have such strong appeal? I think there are several things going on.

First, the histone code idea exists (at least for now) in a kind of limbo: enough biochemical specificity to make it clearly real and a ubiquity that makes is seem important, but sufficiently mysterious that people can imbue it with whatever properties they want. And scientists and non-scientists alike have leapt into this molecular biological sweet spot, using the histone code version of epigenetics as a generic explanation for things they can’t understand, a reason to hope that things they want to be true might really be, and as a difficult to refute, almost quasi-religious, argument for the plausibility of almost any idea linked to heredity.

(On Twitter, Jonathan Eisen presented a revised version of this complaint, with the word “epigenetics” crossed out and “the microbiome” substituted as the hand-waving explanation for everything, and he’s got something there, too!) Another good point is that some of this may well be a reaction to all the genetic determinism of the last twenty years or so: it’s written in your genes, you can’t do a thing about it, everything comes down to DNA, etc. That view is in some ways true, but I’ve always liked E. O. Wilson’s photograph analogy (which unfortunately depends on a vanishing technology). He compared a person’s DNA sequence to a photographic negative, which can be printed in all sorts of ways – long and short exposures, dodging, burning, different sorts of paper and developing solutions, etc, to make all kinds of different-looking prints. But at the same time, none of them contain more information than was there in the original negative.

So I think that Eisen is right – any story that can be told about how your DNA really isn’t your destiny is going to get traction, and perhaps more than it deserves. Lamarckian evolution in particular has always been appealing to people, and I think that goes back to a philosophical/religious part of human psychology. (Note in that passage I quoted from Mukherjee that old-fashioned Darwinian mutation and selection is “glum”). Many people would like to think that the good you do will live after you, perhaps actually imprinted on the heredity you pass on to your descendants, and they may also take it as a warning (or as a grim satisfaction) that the sins of the fathers will be visited on the sons. In general, most of us would probably like to see some karma and cosmic justice meted out, and I think that epigenetics is being dragged in as a substitute for religious feelings. Matthew Arnold and Philip Larkin would have understood perfectly.

25 comments on “Epigenetics For the Spirit”

  1. Anon says:

    Valeant seems to be crashing in freefall again.

  2. McKinsey says:

    The McKinsey consulting epigenetic fingerprint is profound.

  3. pete says:

    Great post, Derek. Especially the 2nd link: Ptashne & Grealy’s critique of fuzzy-minded “Epigenetics”. Interested readers might also checkout Ptashne’s short PNAS opinion piece (free PDF @ ), which was written in response to the ENCODE tornado.

    Side note (rant): in all of these discussions about nature-vs-nurture it seems that Lamarck — y’know, the French biologiste with his long-necked vs short-necked giraffes — tends to get dismissed as a wrong-headed dude who espoused a wrong-headed idea. His ideas were umm..elemental, and were exploring the woods around the tree of natural selection before those 2 English guys got there.

  4. pete says:

    oops.. *Greally*

  5. pete says:

    typo: *Greally*

  6. steve says:

    Except that epigenetic marks can, in fact, be inherited. The way reprogamming works is that methylation marks are converted to hydroxymethyl marks in the germ cells (sperm and egg) and those get diluted and lost during cell division in the embryo. However, some methylation marks escape this process and are indeed transmitted to the next generation. As an example, Mike Skinner exposed mice to vinclozolin (a fungicide) and found that epigenetic changes that occured in first-generation male mice after exposure where carried on through the fourth generation. There are multiple other examples as well. Non-Mendelian inheritance through epigenetic marks is real.

    1. Truth or Truth says:

      I am aware that not all DNA methyl marks are erased. I don’t think the issue is whether these methyl marks are observed in subsequent generations. The question at the heart of the problem is: What is transmitted? In the case of histone marks, here I quote Ptashne (see blog post, mentioned by Derek): “There is no evidence, despite years of research, that nucleosome states can be “copied” for transmission to daughter cells.” That is, you may see the same histone (or DNA methyl) marks in parent and daughter cells, but these are not the heritable unit.

      Separately, here’s Eisen on the issue (again, blog post mentioned by Derek):
      “Ptashne has studied the molecular basis for gene regulation for fifty years. His and Greally’s critique of Mukherjee, or really Allis, is rather technical, and one could quibble about some of the specifics. But his main points are simple and difficult to refute:

      There is essentially no evidence to support the idea that chemical modification of DNA and/or its accompanying proteins is used to encode and transmit information over long periods of time.
      Rather than representing a separate system for storing and conveying information, a wide range of experiments suggests that the primary role of the biochemistry in question is to execute gene expression programs encoded in DNA and read out by a diverse set of proteins known as transcription factors that bind to specific sequences in DNA and regulate the expression of nearby genes.”
      – See more at:

      1. steve says:

        Except again, in Skinner’s experiments they were carried on for at least four generations. If you don’t like that example, look at Agouti mice that have yellow coats and a propensity to become obese and develop cancer because of incomplete erasure of methylation marks at the Avy gene locus.

        1. Bryan says:

          You’re making the same mistake that everyone in the field makes (and which Ptashne et al. rightfully harp on). Evidence for epigenetic inheritance does not mean that these traits are carried by histone or DNA modifications. People somehow started calling these modifications “epigenetic” marks, which has led to the assumption that any observed case of epigenetic inheritance is due to these “epigenetic” marks. Yes, in plants, there are some well characterized cases of epialleles where DNA methylation changes can be inherited across generations without underlying changes in the DNA sequence, but their DNA methylations systems are much different from those found in animals. Furthermore, many of these methylation events are regulated by small RNAs, and in both plants and animals, many small RNAs are transmitted from both parents to their offspring, which provides an alternative explanation for some of the cases of transgenerational epigenetic inheritance.

          Changes to histone modifications are certainly correlated with changes in gene expression, but the issue that causes arguments at conferences is the question of causation. Do the small number of DNA methylations and histone modifications transmitted from parent to offspring cause epigenetic inheritance? Or is it due to inheritance of other factors like transcription factors or RNAs (and the changes to histone and DNA modifications are only secondary results of these inherited proteins/RNAs)?

          1. Steve says:

            I think the agouti case is pretty clearcut that it’s epigenetic since its effects are differential DNA methylation. With regard to your alternative hypothesis – that its due to cytoplasmic factors like RNAs – it’s pretty hard to argue that. For example, another case of epigenetic inheritance, the kinky tail of the Axin-fused mouse, is transmitted paternally. Sperm contribute very little in the way of cytoplasmic factors to the embryo so it’s pretty hard to understand how RNA could lead to a change in DNA methylation in these mice that’s transmitted from generation to generation. Any hypothetical RNA would be quickly diluted out.

          2. Bryan says:

            Is DNA methylation the cause or the effect of the transgenerational inheritance observed in those experiments? This question is very difficult to answer. DNA and histone modification are certainly important in the regulatory mechanism, but too many times scientists in this field assume that correlation = causation.

            As for transgenerational inheritance of RNA, experiments in C. elegans suggest these can be passed transgenerationally (see for example Rechavi et al. 2011 Cell 147: 1248, Rechavi et al. 2014 Cell 158: 277). Whether these same mechanisms operate in mammals is unclear. However, small RNAs inherited through sperm have been demonstrated to affect the phenotype of offspring intergenerationally in mice (see for example, Sharma et al. 2016 Science 351: 391, Chen et al. 2016 Science 351: 397).

          3. Truth or Truth says:

            Thanks Bryan. I think you are dead on.

  7. Truth or Truth says:

    Thanks for covering this story. The Emperor of All Maladies was an incredible book, a real tour de force–I am hoping that this recent article was just a gross oversight that will be forgiven in time.

    I wanted to clarify one point in your post.

    “Particularly upsetting to many biologists was the way the article never even mentioned the phrase “transcription factors”, when the main function of all these epigenetic markers is, in fact, to harness and guide such proteins.”

    Most biologists who are upset view this relationship the other way around: It is the *transcription factors* that dictate the placement of these so-called epigenetic marks, not the other way around (quotes from the blog link below). Remember, the enzymes that place these marks do not have any specificity for a particular genomic location. They are directed to a locus be a sequence-specific DNA binding protein.

    Quoting Ptashne:
    “The Yamanaka experiment, in fact, showed the opposite: that you can change cell identity by expressing certain DNA-binding proteins that bind to and activate specific genes. Any changes in chromatin organization—presumably the “epigenetic marks” referred to, given the context of the entire piece—found during this process are the result of the activities of the DNA-binding regulatory proteins Yamanaka used. ”

    Quoting Steve Henikoff (from part 1 of the blog):
    “Mukherjee seemed not to realize that transcription factors occupy the top of the hierarchy of epigenetic information, that this has been widely accepted in the broader chromatin field, and that histone modifications at most act as cogs in the machinery that enforces the often complex programs specified by the binding of transcription factors.”

    1. steve says:

      See my earlier comment in reply to Bryan. The argument about transcription factors is a bit silly, just another chicken-and-egg argument best left to beer halls. The bottom line is that methylation can change gene expression and can be transmitted from generation to generation; epigenetic transmission of traits is therefore real no matter how some may try to argue it away.

  8. NJBiologist says:

    “That view is in some ways true, but I’ve always liked E. O. Wilson’s photograph analogy (which unfortunately depends on a vanishing technology).”

    Just substitute RAW files for negatives, and Photoshop for all the analog manipulations. The analogy still works.

  9. steve says:

    You might also want to point out that Mukherjee just launched a biotech together with PureTech based on some CAR-T technology he developed. It’ll be interesting to see if he really has something novel; if so, he’s quite impressive regardless of the opinions about his article on epigenetics.

  10. steve says:

    I’m not trying to dominate this thread but I just re-read the Michael Eisen quote above that this is an “obscure biochemical process”. In fact, the number of epigenetic changes in cancer dwarfs that of mutations by at least an order of magnitude. Changes in epigenetics alone (without mutation) can cause cancer as shown by the Baylor group working with p16. HDAC inhibitors as well as drugs targeting histone methyltransferases, demethylases and bromodomains are now mainstream cancer drugs and are also used to treat cardiovascular disease, psychiatric diseases, etc. So much for obscurity.

  11. Garrett Wollman says:

    Biologists love to rag on Lamarckianism, for good reason (he was wrong about how biological evolution works, and Darwin and Wallace were [mostly] right), but it’s worth keeping in mind that some things actually do evolve in a Lamarckian way, in particular culture (sensu lato), which is something professional writers might reasonably be expected to have a particular affinity for.

  12. David Borhani says:

    Prompted by this interesting discussion, I re-read Mukherjee’s The New Yorker article online instead of print (I’m a Luddite), so I could search in it for text quoted (or suggested) in Mukherjee’s response to critics in the interview with Matthew Herper.

    I think Mukherjee’s response further obfuscates rather than clarifies the many errors in the original New Yorker article.

    Mukherjee states to Herper:

    Point two is that…the idea that transcription factors are denied their role in the piece is wrong. There are four or five references to transcription factor genes being turned on and off in cells that are explicitly in the piece. There’s an incredible sensitivity to that idea. That’s why these things were put in, and if you look at the piece carefully, you’ll see that every time we talk about histone modification, we talk about transcription factors. I’ll give you one example and you can look for the rest.

    Number one, one example I tell you is that in discussing one sentence says, “Genes are turned on and off in response to these cues and epigenetic marks are laid subsequently or later, whatever they might be.” That’s just one example of how we make it very clear, because we knew the field, I knew the field very well. Genes are turned on and off and transcription factors turn on and off, and epigenetic match factors are secondary.

    Well, the word “transcription” (or miRNA) never appears in the article; “epigenetic” appears 36 times. And “very clear”? I disagree. What Mukherjee actually wrote in the New Yorker was:

    Chance events—injuries, infections, infatuations; the haunting trill of that particular nocturne—impinge on one twin and not on the other. Genes are turned on and off in response to these events, as epigenetic marks are gradually layered above genes, etching the genome with its own scars, calluses, and freckles. Prospero, raging against the deformed Caliban in “The Tempest,” describes him as “a devil, a born devil, on whose nature/Nurture can never stick.” Caliban is destined to remain a genetic automaton, a windup ghoul—vastly more pathetic than anything human. He experiences the world, but he has no capacity to be changed by it; he has a genome that lacks an epigenome.

    Does that second sentence—“Genes are turned on and off in response to these events, as epigenetic marks are gradually layered above genes…” clarify or obscure that ghostly, never-mentioned-by-name transcription factors, not “epigenetic” marks, are the main, and initial, actors here? Any possible confusion in favor of the spectre-like transcription factors is erased by the last sentence, “He experiences the world, but he has no capacity to be changed by it; he has a genome that lacks an epigenome.”

    Further, reiterating a point made by Ptashne and Greally, Mukerjee writes (in the original) “Had Allis started his experiments in the nineteen-eighties trying to pin down words like “identity” and “memory,” he might have found himself lost in a maze of metaphysics.” and “If he [Allis] could identify the molecular switches that maintain one state, or toggle between the two states, he might be able to identify the mechanism responsible for cellular memory.” Why didn’t Mukherjee at least mention at this point that the fundamental positive feedback mechanism—“A Genetic Switch” (book published in 1986)—had already been figured out, for bacteriophage lambda, by Ptashne and coworkers? I guess it would have interfered with his “narrative”. And space is really not an issue in a New Yorker article already of such length, as Mukherjee claimed to Herper.

    All of this is a bit surprising and rather unfortunate. Mukherjee is a great writer, and he has a true gift for explaining science to the uninitiated. But, I guess in genetics (cf. his forthcoming book, of which the New Yorker article is an excerpt) he’s just out of his league.

    1. tangent says:

      “He experiences the world, but he has no capacity to be changed by it; he has a genome that lacks an epigenome.”

      Yeah, this surely looks like motivated reification of this underspecified “epigenome”, which seems to be used for “the capacity to be changed by the world in a lasting way”.

      People want an alternative to genetic determinism (and that’s especially understandable if they’ve been exposed to pop-sci “found the gene for” genetics). Can organisms change? Well no shit they can, but people feel like it adds something to point to a fancy biochemical mechanism.

      It’s like the power of brains. Can human minds be lastingly affected by experiences? Well no shit they can. But people impress other people by pointing to “a detectable structural change in the brain linked to [whatever it is]”, like that makes the effect itself more real.

  13. tangent says:

    “But at the same time, none of them contain more information than was there in the original negative.”

    That’s mildly not the case for photo prints versus negatives, and seriously not the case for the organism versus its genome, though. Unless some odd sense of “information” is being used. This analogy just puts up a punching bag for epigenetic woo artists to go after, because it’s a bad analogy.

    The organism has everything that makes identical twins different people, all the effects of a lifetime in the world. Memory, to state the obvious.

    Look at it quantitatively: a human genome is a few gigapairs, taking no distinction between coding and non-, so we’re talking order of a 1e+9 bytes of genetic information. The brain is order of 100e+9 neurons making 100e+12 synapses, which each carry state. Sure, some may be more or less foreordained by the genome. Sure, there’s redundancy that makes the true information content lower. But bottom line, the phenotype is far more information than the genome. Very much unlike a print of a negative.

  14. Ted Cruise says:


  15. chiz says:

    MZ twins are not genetically identical. They are much much more similar than DZ (or SZ) twins but they are not identical. This is well established empirically now and it seems strange that so many people, especially psychologists, haven’t heard about it and keep looking to other things, such as epigenetics, to explain differences between MZ twins.

  16. Argon says:

    Yeah, unfortunately, the term, ‘epigenetics’ is replacing ‘quantum’ in quack medicine and science. So of course ‘quantum epigenetics’ is even cooler! I googled the latter and found at the top of the search (sigh) a breathless scientific article titled “Epigenetics: Biology’s Quantum Mechanics” (

    An article in the Guardian described this phenomenon last year, “Beware the pseudo gene genies”:–darwin-adam-rutherford

    Sadly, the reality, ‘Epigenetics: Yet another regulatory mechanism’, doesn’t sound as sexy.

Comments are closed.